Usage of machine learning for the separation of electroweak and strong $Z_{\gamma}$ production at the LHC experiments

Separation of electroweak component from strong component of associated Zγ production on hadron colliders is a very challenging task due to identical final states of such processes. The only difference is the origin of two leading jets in these two processes. Rectangular cuts on jet kinematic variables from ATLAS/CMS 8 TeV Zγ experimental analyses were improved using machine learning techniques. New selection variables were also tested. The expected significance of separation for LHC experiments conditions at the second datataking period (Run2) and 120 fb−1 amount of data reaches more than 5σ. Future experimental observation of electroweak Zγ production can also lead to the observation physics beyond Standard Model.